Merge pull request #22 from badboy/regression-21/backslash-escape

.github/workflows/tests.yml

@@ -23,12 +23,6 @@ jobs:
         toolchain: ${{ matrix.rust }}
         override: true
 
-    - name: check
-      uses: actions-rs/cargo@v1
-      with:
-        command: check
-        args: --all
-
     - name: tests
       uses: actions-rs/cargo@v1
       with:

Cargo.toml

@@ -10,13 +10,12 @@ edition = "2018"
 
 [dependencies]
 mdbook = "0.4.10"
-pulldown-cmark = "0.8.0"
-pulldown-cmark-to-cmark = "6.0.2"
-env_logger = "0.8.4"
+pulldown-cmark = "0.9.1"
 log = "0.4.11"
 clap = "2.33.3"
 serde_json = "1.0.57"
 toml = "0.5.6"
 
 [dev-dependencies]
-pretty_assertions = "0.6.1"
+pretty_assertions = "1.0.0"
+env_logger = "0.9.0"

src/lib.rs

@@ -8,7 +8,6 @@ use mdbook::errors::{Error, Result};
 use mdbook::preprocess::{Preprocessor, PreprocessorContext};
 use pulldown_cmark::Tag::*;
 use pulldown_cmark::{Event, Options, Parser};
-use pulldown_cmark_to_cmark::{cmark_with_options, Options as COptions};
 use toml::value::Table;
 
 pub struct Toc;
@@ -110,6 +109,7 @@ fn build_toc(toc: &[(u32, String, String)]) -> String {
     let mut toc_iter = toc.iter().peekable();
 
     // Start from the level of the first header.
+    let min_level = toc.iter().map(|(lvl, _, _)| *lvl).min().unwrap_or(1);
     let mut last_lower = match toc_iter.peek() {
         Some((lvl, _, _)) => *lvl,
         None => 0,
@@ -127,7 +127,7 @@ fn build_toc(toc: &[(u32, String, String)]) -> String {
     });
 
     for (level, name, slug) in toc {
-        let width = 2 * (level - 1) as usize;
+        let width = 2 * (level - min_level) as usize;
         writeln!(result, "{1:0$}* [{2}](#{3})", width, "", name, slug).unwrap();
     }
 
@@ -135,7 +135,6 @@ fn build_toc(toc: &[(u32, String, String)]) -> String {
 }
 
 fn add_toc(content: &str, cfg: &Config) -> Result<String> {
-    let mut buf = String::with_capacity(content.len());
     let mut toc_found = false;
 
     let mut toc_content = vec![];
@@ -150,40 +149,41 @@ fn add_toc(content: &str, cfg: &Config) -> Result<String> {
     opts.insert(Options::ENABLE_TASKLISTS);
 
     let mark: Vec<Event> = Parser::new(&cfg.marker).collect();
-    let mut mark_start = -1;
+    let mut mark_start = None;
+    let mut mark_end = 0..0;
     let mut mark_loc = 0;
-    let mut c = -1;
 
-    for e in Parser::new_ext(&content, opts) {
-        c += 1;
-        log::trace!("Event: {:?}", e);
+    for (e, span) in Parser::new_ext(&content, opts).into_offset_iter() {
+        log::trace!("Event: {:?} (span: {:?})", e, span);
         if !toc_found {
             log::trace!(
-                "TOC not found yet. Location: {}, Start: {}",
+                "TOC not found yet. Location: {}, Start: {:?}",
                 mark_loc,
                 mark_start
             );
             if e == mark[mark_loc] {
-                if mark_start == -1 {
-                    mark_start = c;
+                if mark_start.is_none() {
+                    mark_start = Some(span.clone());
                 }
                 mark_loc += 1;
                 if mark_loc >= mark.len() {
+                    mark_end = span;
                     toc_found = true
                 }
             } else if mark_loc > 0 {
                 mark_loc = 0;
-                mark_start = -1;
+                mark_start = None;
             } else {
                 continue;
             }
         }
 
-        if let Event::Start(Heading(lvl)) = e {
-            current_header_level = Some(lvl);
+        if let Event::Start(Heading(lvl, fragment, classes)) = e {
+            log::trace!("Header(lvl={lvl}, fragment={fragment:?}, classes={classes:?})");
+            current_header_level = Some(lvl as u32);
             continue;
         }
-        if let Event::End(Heading(_)) = e {
+        if let Event::End(Heading(..)) = e {
             // Skip if this header is nested too deeply.
             if let Some(level) = current_header_level.take() {
                 let header = current_header.clone();
@@ -219,29 +219,30 @@ fn add_toc(content: &str, cfg: &Config) -> Result<String> {
 
     let toc = build_toc(&toc_content);
     log::trace!("Built TOC: {:?}", toc);
-    let toc_events = Parser::new(&toc).collect::<Vec<_>>();
+    log::trace!("toc_found={toc_found} mark_start={mark_start:?} mark_end={mark_end:?}");
 
-    let mut c = -1;
-    let events = Parser::new_ext(&content, opts)
-        .map(|e| {
-            c += 1;
-            if toc_found && c > mark_start && c < mark_start + (mark.len() as i32) {
-                vec![]
-            } else if toc_found && c == mark_start {
-                toc_events.clone()
+    let content = if toc_found {
+        let mark_start = mark_start.unwrap();
+        let content_before_toc = &content[0..mark_start.start];
+        let content_after_toc = &content[mark_end.end..];
+        log::trace!("content_before_toc={:?}", content_before_toc);
+        log::trace!("content_after_toc={:?}", content_after_toc);
+        // Multiline markers might have consumed trailing newlines,
+        // we ensure there's always one before the content.
+        let extra = if content_after_toc.as_bytes()[0] == b'\n' {
+            ""
         } else {
-                vec![e]
-            }
-        })
-        .flatten();
-
-    let opts = COptions {
-        newlines_after_codeblock: 1,
-        ..Default::default()
+            "\n"
         };
-    cmark_with_options(events, &mut buf, None, opts)
-        .map(|_| buf)
-        .map_err(|err| Error::msg(format!("Markdown serialization failed: {}", err)))
+        format!(
+            "{}{}{}{}",
+            content_before_toc, toc, extra, content_after_toc
+        )
+    } else {
+        content.to_string()
+    };
+
+    Ok(content)
 }
 
 impl Toc {

tests/adds_toc.out.md

@@ -18,3 +18,4 @@
 ## Header 2.2
 
 ### Header 2.2.1
+

tests/backslash_escapes.in.md

@@ -0,0 +1,9 @@
+\*not emphasized*  
+\<br/> not a tag  
+\[not a link](/foo)  
+\`not code`  
+\* not a list  
+\# not a heading  
+\[foo]: /url "not a reference"  
+\&ouml; not a character entity  
+1\. not a list  

tests/backslash_escapes.out.md

@@ -0,0 +1,9 @@
+\*not emphasized*  
+\<br/> not a tag  
+\[not a link](/foo)  
+\`not code`  
+\* not a list  
+\# not a heading  
+\[foo]: /url "not a reference"  
+\&ouml; not a character entity  
+1\. not a list  

tests/handles_inline_code.in.md

@@ -13,5 +13,3 @@
 ##### Header 1.1.1.1.1
 
 # Another header `with inline` code
-
-

tests/it.rs

@@ -58,7 +58,7 @@ macro_rules! assert_toc {
         let chapter = Chapter::from_content(content);
         let result = Toc::add_toc(&chapter, &config);
         match result {
-            Ok(result) => assert_eq!(expected.trim_end(), result),
+            Ok(result) => assert_eq!(expected, result),
             Err(e) => panic!("{} failed. Error: {}", $name, e),
         }
     };
@@ -114,7 +114,7 @@ fn unique_slugs() {
 
 #[test]
 fn add_toc_with_github_marker() {
-    let marker = "* auto-gen TOC:\n{:toc}".to_owned();
+    let marker = "* auto-gen TOC:\n{:toc}\n".to_owned();
     assert_toc!("github_marker", with_marker(marker));
 }
 
@@ -150,3 +150,10 @@ fn similar_heading_different_casing() {
 fn tables_with_html() {
     assert_toc!("tables_with_html");
 }
+
+#[test]
+fn backslash_escapes() {
+    // Regression test #21
+    // Backslash-escaped elements should still be escaped.
+    assert_toc!("backslash_escapes");
+}

tests/multi_header_linear.out.md

@@ -7,14 +7,9 @@
   * [Level 1.2.1](#level-121)
 
 ## Level 1.1
-
 ### Level 1.1.1
-
 ### Level 1.1.2
-
 ## Level 1.2
-
 ### Level 1.2.1
 
 text
-

tests/tables_untouched.out.md

@@ -1,5 +1,6 @@
 # Heading
 
 | Head 1 | Head 2 |
-|------|------|
+|--------|--------|
 | Row 1  | Row 2  |
+

tests/tables_with_html.out.md

@@ -1,5 +1,5 @@
 # Heading
 
 | Head 1 | Head 2 |
-|------|------|
+|--------|--------|
 | <span>Row 1</span> | Row 2  |

tests/with_inline_code.in.md

@@ -9,3 +9,6 @@
 # Header 2
 
 ## Header 2.1
+
+When code or an identifier must appear in a message or label, it should be
+surrounded with backticks: `` `foo.bar` ``.

tests/with_inline_code.out.md

@@ -12,3 +12,6 @@
 # Header 2
 
 ## Header 2.1
+
+When code or an identifier must appear in a message or label, it should be
+surrounded with backticks: `` `foo.bar` ``.